Transform domain steganography detection using fuzzy inference systems

Steganography is a means of data hiding in images for covert transmission. Though steganography aims at transmitting images without visual degradation or changes for a naked observer, it cannot dispense with altering spatial and transform level details in order to embed the data [N. F. Jhonson (2002), A. Westfeld, et al., (2000)], Even though these alterations may not be captured by visual observation, they do manifest themselves for detailed analysis. In this paper we analyze a fuzzy logic based technique that could be applied for detaining images with steganography. The images we are looking at are JPEG images and the data is hidden on the compressed image's discrete cosine transform (DCT) coefficients. A similar technique was employed in our earlier work for detection of spatial domain steganographic images [A. Westfeld, et al., (2000)]. The main reason we extend the same idea in this paper is to maintain consistency with established techniques and in achieving universal stego-detection. Universal stego-detection is any process that would detect steganography irrespective of whether done in the spatial domain or transform domain [R. Chandramouli et al., (2001)].